DCGAN is initialized with random weights, so a random code plugged into your network would crank out a completely random graphic. On the other hand, while you may think, the network has a lot of parameters that we will tweak, and the target is to find a location of those parameters which makes samples produced from random codes look like the teachi